Nanoscale - Harvard University

Nanoscale

PAPER

aSchool of Engineering and Applied Scie

University, Cambridge, MA 02138, USA. TelbETH Zurich, Department of Materials, Labo

Zurich, Switzerland. E-mail: [email protected]

4633 6055cCenter for Systems Biology, Massachusetts GdDepartment of Bioengineering, School of En

of Pennsylvania, Philadelphia, Pennsylvania

† Electronic supplementary informationmicrouidic device; measurements of deon the temperature of benzyl alcohoproperties of common solvents; a quasi-HRTEM and SEAD of nanoparticles10.1039/c3nr00500c

Cite this: Nanoscale, 2013, 5, 5468

Received 28th January 2013Accepted 4th April 2013

DOI: 10.1039/c3nr00500c

www.rsc.org/nanoscale

5468 | Nanoscale, 2013, 5, 5468–547

Microwave dielectric heating of non-aqueous dropletsin a microfluidic device for nanoparticle synthesis†

Dorota Koziej,*ab Caspar Floryan,a Ralph A. Sperling,a Allen J. Ehrlicher,a

David Issadore,cd Robert Westervelta and David A. Weitza

We describe a microfluidic device with an integrated microwave heater specifically designed to

dielectrically heat non-aqueous droplets using time-varying electrical fields with the frequency range

between 700 and 900 MHz. The precise control of frequency, power, temperature and duration of the

applied field opens up new vistas for experiments not attainable by conventional microwave heating.

We use a non-contact temperature measurement system based on fluorescence to directly determine

the temperature inside a single droplet. The maximum temperature achieved of the droplets is 50 �C in

15 ms which represents an increase of about 25 �C above the base temperature of the continuous

phase. In addition we use an infrared camera to monitor the thermal characteristics of the device

allowing us to ensure that heating is exclusively due to the dielectric heating and not due to other

effects like non-dielectric losses due to electrode or contact imperfection. This is crucial for illustrating

the potential of dielectric heating of benzyl alcohol droplets for the synthesis of metal oxides. We

demonstrate the utility of this technology for metal oxide nanoparticle synthesis, achieving

crystallization of tungsten oxide nanoparticles and remarkable microstructure, with a reaction time of

64 ms, a substantial improvement over conventional heating methods.

Introduction

The properties of nanoparticles differ from those of their bulkcounterparts due to the connement effect caused by their size,structure, and shape.1 To optimize these properties for specicapplications it is useful to carry out synthetic reactions withsmall volume and low cost. Microuidic reactors are valuabletools for this application, due to their ability to rapidly dosereagents and to create homogenous mixtures on the scale ofmicrons.2–7 Microuidic systems can be used either in contin-uous ow or segmented ow through the use of droplets. Theuse of segmented ow reactors, in comparison with theircontinuous counterparts, allows nanoliter volumes of reactionsolutions to be independently controlled with precise reaction

nces, Department of Physics, Harvard

: +1 61 7496 9788

ratory for Multifunctional Materials, 8093

hz.ch; Fax: +41 14 633 15 45; Tel: +41 14

eneral Hospital, Boston, MA 02114, USA

gineering and Applied Science, University

, USA

(ESI) available: Optical images of thependence of the microwave frequencyl droplets; a summary of dielectricstatic electrical eld simulation; TEM,

washed with ethanol. See DOI:

5

conditions, such as heat, mass transfer rates, and tempera-ture.2–7 The ability to separate and control the stages of nucle-ation and growth is a unique aspect of microuidic synthesis.These properties found broad application in fabrication ofmetal,8 III–VI semiconductor5 and polymeric9 nanoparticleswith dened size and structure. Recent developments have beendirected toward scaling out of the processes by adding reactorsworking in parallel.10,11 Despite enormous progress in usingmicrouidic reactors for the synthesis of nanoparticles, thecrystallinity of metal oxide nanoparticles has only beendemonstrated in systems where the reactants are maintained athigh temperatures for at least 1 to 30 minutes. This can beaccomplished through the use of a resistive heater, an oil bathor an oven, which heats either the entire device, or just thecollection tube.4,12 An alternate means of efficiently increasingthe temperature is to heat the sample alone. This can be donethrough the use of microwaves, which selectively and directlyheat the sample itself.12–14 Microwave heating can signicantlyincrease the rate of synthesis of metal oxide nanoparticles in thebulk, reducing the reaction time from hours to minutes.15

Recently, we have shown that microwave heating can beintegrated into droplet based microuidic chips, enablingrapid (<30 ms) heating to temperatures 30 �C above the basetemperature.16 In this previous study, we heated water atf ¼ 2.4 GHz, utilizing its large dielectric resonance loss atmicrowave frequencies. Several novel device geometries andmaterials have been employed to fabricate the efficient

This journal is ª The Royal Society of Chemistry 2013

Paper Nanoscale

transmission of microwave power to samples on microuidicchips.17,18 Because the research on microuidic-microwavereactors has been driven by its potential application for DNAamplication or cell lysis, the state-of-the-art devices are dedi-cated to heating of water in the frequency range between 2.45and 15 GHz. However, water is not a suitable solvent for thelow temperature synthesis of many inorganic metal oxides incrystalline states. Instead solvents such as benzyl alcohol areutilized, since in the bulk they enable both rapid microwavesynthesis and efficient crystallization at relatively low reactiontemperatures.15 Despite the widespread use of microwaveheating in the synthesis of nanoparticles, microwave heatinghas never been done with a microuidic device and theadvantages of this technique have not been explored.

In this paper we describe a microuidic dielectric-heatingdevice, operating at 700–900 MHz, which is capable of preciselyheating the non-aqueous solvents used for nanoparticlessynthesis, for instance benzyl alcohol, n-butanol and ethyleneglycol.19–23 Label-free IR temperature imaging provides quanti-tative information about the dielectric heating of the benzylalcohol droplets, heat transfer from the droplets, and non-dielectric losses. Additionally we measure with high temporalresolution the dielectric heating of the benzyl alcohol dropletsby uorescence imaging. To demonstrate the utility of thisapproach, we synthesize crystalline tungsten oxide nano-particles in benzyl alcohol using our chip. Typically tungstenoxide synthesis leads to a WO3$nH2O product in the form ofplatelets independent of the heating method.24,25 Tungstenoxides nanoparticles have been widely used materials forapplication in gas sensing, water splitting, electrochromicwindows. In addition, unconventional applications of tungstenoxide nanoparticles are for high Tc superconductors, eldemission displays and optical recording devices.26–29 Therefore,we choose this material to show the feasibility of microwavesynthesis on a microuidic chip.

Materials and methodsDevice

The microuidic device is fabricated using standard solithography. The PDMS channels are 50 mm high. The devicewas plasma bonded to a thin glass slide (0.13mm), the channelswere functionalized with Aquapel and baked for 20 minutes at60 �C. The electrodes are fabricated by applying a low-meltingsolder ll technique previously described.30 The electronics togenerate the microwave power at 700–900 MHz is assembledsimilar to the electronics for 3 GHz frequency.16 Themicrowavesare generated with a voltage-controlled oscillator (ZX95-1200W+, Mini-Circuits) and are amplied with a power ampli-er (ZHL-211-8 Mini-Circuits). The quasi-static simulation ofthe electrical eld between the electrodes is performed withAnso, Maxwell soware.

Calibration of temperature measurements

We use rhodamine B (RhB) and rhodamine 110 (Rh110) at aconcentration of 0.1 mmol and measure the emission spectra


on a Leica TCS SP5 confocal microscope equipped with aheating stage (Warner Instruments). The dyes are excited at488 nm and the emission is examined by scanning in 5 nmsteps between 500 and 650 nm at different temperatures. Thetwo dyes have different temperature-dependent spectra whenexcited at 488 nm allowing the individual signal to be separatelymeasured and the temperature of the sample determined bycalibration.

Single droplet temperature measurement set-up

The detection setup consists of a 50 mM, 488 nm cw lasercoupled into the backport of a Motic AE-31 uorescencemicroscope. The laser is focused by a 40�, NA 0.85 objectiveinto the ow channel of the microuidic device, which is heatedby microwaves. The emitted uorescence is collected by thesame objective and passed through a series of dichroic mirrorsat the back of the microscope. Photomultiplier tubes behind thebandpass lter detect the light at 536/40, data acquisition iscarried out on a PC with a FPGA card fromNational Instrumentsand a LabView program.

Thermal IR imaging

The temperature prole of the device is measured with an IRcamera SC 5600 FLIR equipped with an InSb detector using aframe rate 100 Hz. The 0.13 mm thick glass slide is partiallytransparent to infrared; therefore we cannot image completelythrough the glass and measure temperature directly in thechannel. In the detector spectral range (3–5 mm) the trans-mittance of glass lies between 90 and 40%.

Chemicals

Anhydrous benzyl alcohol ($99%), tungsten hexachloride($99.9%), rhodamine B (97%), and rhodamine 110 ($99.0%),oxalyl chloride (98%), poly(propylene glycol)-block-poly(ethyleneglycol)-block-poly(propylene glycol) bis(2-aminopropyl ether),anhydrous dichloromethane (99.8%) were supplied by Sigma-Aldrich, Fluorinert Electronic Liquid FC-40 (FC-40 oil) and HFE-7100 by 3 M, polyethylene oxide, Mw 100 000 by Alfa Aesar,Krytox 157 FSH by Miller-Stephenson.

All chemicals were used without further purication. SU83050 was purchased from MicroChem, polydimethylsiloxane(PDMS) prepolymer and curing agent – Sylgard 184 – were fromEssex Brownell, indium alloy Ind19 (52In, 32.5 Bi, 16.5 Sn)0.0200 0 diameter was from Indium Corporation, Aquapel wasfrom Pittsburgh Glass Works, LLC.

Synthesis in a microuidic device

Tungsten hexachloride (20 mg) was added, in an oxygen- andwater-free atmosphere, to anhydrous benzyl alcohol (3 ml). Thelight blue solution was directly transferred to a glass syringe.The surfactant to stabilize benzyl alcohol droplets in FC-40 oilwas synthesized according to the published procedure.31 Then,1 wt% of surfactant was dissolved in FC-40 and the resultingsolution was ltered and transferred to a glass syringe. Theliquids were injected into the channels using syringe pumps

Nanoscale, 2013, 5, 5468–5475 | 5469

Nanoscale Paper

from Harvard Apparatus. The experiments were performed at aconstant ow rate of benzyl alcohol at 20 ml h�1 and the oil at120 ml h�1, unless otherwise stated. Aer heating the droplets ina microuidic-microwave device the solution turned fromtransparent bluish to slightly yellowish. In the control experi-ments, the reaction solution was heated in a CEM Explorerlaboratory microwave reactor in a 10 ml vessel (7.5 ml reactionsolution) at 60 �C for 1–30minutes and at 120 �C for 10minutes.At 60 �C (1 to 15 minutes) of reaction we did not observeformation of precipitate even aer the solution was centrifugedfor several hours at a speed of 15 000 rpm.

Synthesis of poly(peruoropropylene glycol)–PEG

Krytox 157 FSH is a peruorinated polyether with a carboxylicgroup at one end. The carboxylic group is converted to the acidchloride by reacting it with an excess of oxalyl chloride. Aerevaporating off the excess of oxalyl chloride, a diamine PEG isadded. The acid chloride of the uorinated block reacts with theamino groups of the PEG, forming di- and triblock molecules.Unreacted PEG is separated by centrifugation; evaporation ofthe solvent yields the surfactant that can be used withoutfurther purication.

Material characterization

Transmission Electron Microscopy (TEM) and high-resolutionTransmission Electron Microscopy (HR-TEM) investigationswere performed on a Philips Tecnai F30 at 300 kV. ScanningElectron Microscopy (SEM) studies were performed on a ZeissFESEM Supra55VP at 5 kV. The droplets drip directly from thecollection tube of the PDMS device onto carbon-coated coppergrids (TEM) or the alumina stub (SEM) were immediatelyvacuum dried. X-ray powder diffraction (XRD) was measuredusing a Panalytical XPert Pro Diffractometer (45 kV, 40 A)equipped with an X'cellerator detector and xed slits. Therepeated-scans at 2 theta from 10 to 60� weremerged to improvethe signal-to-noise ratio.

Results and discussionDevice for microwave heating of non-aqueous solvents

Ourdroplet basedmicro-reactorwas fabricated fromPDMSusingso lithography and consists of a droplet maker followed by amicrowaveheater as schematically illustrated in Fig. 1 and shown

Fig. 1 Schematic of the microfluidic device. The flow-focusing droplet maker isfollowed by a microwave heater. The benzyl alcohol droplets are rapidly heatedusing microwaves of 862 MHz frequency (7.5 V tuning voltage). Each droplet is amicro-reactor for ultra-fast nanoparticle synthesis; one example is tungsten oxide.

5470 | Nanoscale, 2013, 5, 5468–5475

in ESI Fig. S1,† The droplet maker is in the ow focusing geom-etry, with inlets that are each 20 mm wide and a nozzle thatsmoothlywidens from20mmto50mm.Microwavepower is locallydelivered by indium-alloy electrodes that are directly integratedinto the microuidic device16,30 and are situated 0.5 cm down-stream from the droplet maker. The frequency of the rotationalrelaxation process of benzyl alcohol molecules is approximately850 MHz.20 In order to efficiently generate heat the frequency ofthe heater should approximately match with that of benzylalcohol. At 2.45 GHz the molecules of benzyl alcohol are not ableto follow theelectricaleldoscillationandbehave like a less-polarsolvent. Thereforeadevicedesigned toheatwaterwithmicrowaveheating using frequency 2.45 GHz (400 mW) is not efficient forheating of benzyl alcohol. Conventional laboratory microwaveovens compensate for the low energy absorption of non-aqueoussolvents at 2.45 GHz by using high output power. Instead for ourexperiments, we construct an inexpensive microwave source thatcan supply up to 800mW in the frequency range between 700 and900 MHz. Although, we test this device at different frequencieswith benzyl alcohol (see ESI Fig. S2†), it can also be used withoutfurther modication with other common non-aqueous solventsfor nanoparticle synthesis.21–23,32 For example 1-hexanol, 1-butanol, 2-propanol, glycerol and diethylene glycol all havehigher relaxation times (s) than that of water and therefore theirmaximal dissipation factors (d) are below 2.45 GHz (Table S1,ESI†). The microwave power is controlled by varying the peak-to-peak voltage of the amplier, and the frequency is controlled by avoltage controlled oscillator. We match the impedance of theelectrodes to that of the source and optimize the electrodesgeometry to ensure efficient delivery of microwave power to thedroplets; we model the quasi-static electrical eld to help opti-mize the device (see ESI, Fig. S3†). As a uid we use FC-40 3M oilwith 1wt%Krytox–PEG surfactant31 for the continuous phase andbenzyl alcohol for the dispersed phase.

Temperature within the benzyl alcohol droplets

To determine the temperature within the droplets we measurethe uorescence intensity of dyes dissolved in the benzylalcohol. We rst calibrate the temperature dependence of theuorescence intensity of rhodamine B and 110, since little isknown about their behavior in benzyl alcohol.33,34 Interestingly,the uorescence intensity of each dye decreases linearly withtemperature; this is unlike the behavior in water where onlyRhB is temperature dependent.35,36 The measured temperaturesensitivities of RhB at 575 nm and Rh110 at 530 nm are verysimilar and are 1.75 � 0.08%/�C and 1.74 � 0.08%/�C, respec-tively (see ESI Fig. 4 and 5†). Further, we measure the uores-cence intensity from both dyes in droplets of benzyl alcohol inthe microuidic chip, and use this to determine the tempera-ture within the droplets when they are heated with the micro-waves. Fluorescence is excited by a 488 nm, 50 mW laser that isfocused on the droplets when they are in the ow channel,between the electrodes used for the microwave heating. Theuorescence emitted from each individual droplet is collectedby a microscope objective and guided through a series ofdichroic mirrors to three photomultiplier tubes. We record the


Paper Nanoscale

uorescence intensity of many droplets as a function of thepeak voltage applied to the microwave heater, and determinethe average over one thousand equally sized droplets. To ensurethe accuracy of the results we pick the equally sized dropletsfrom the whole population of droplets by setting the lower limitof uorescence detection and the duration of droplets, asshown in the ESI, Fig. 6.†

In contrast to long-exposure uorescence imaging of manydrops traveling through the microwave heater,37–39 this methoddoes not require sophisticated background extraction andperfectly equally sized drops. The temperature rises linearly withmicrowavepower, which is proportional to the square of thepeakvoltage applied to the electrodes, as shown in Fig. 2. Themaximum temperature of the droplets is 50 �C which representsan increase of about 25 �C in 15 ms. Interestingly, when wecompare the temperature of droplets different in size, we candistinguish two regimes; for droplets smaller than the size of thechannel, the slope of the temperature increase with microwavepower is steeper and the maximum temperature is higher thanthat for the larger droplets. Because the larger droplets aredeformed by the walls and therefore in close proximity to thePDMS, heat is more efficiently conducted to the bulk of thedevice; in contrast the carrier oil has a low thermal conductivity(kFC-40 0.065 W m�1 C�1)40 which, in the case of smaller drop-lets, effectively prevents dissipation of heat from the dropletsinto the device. The standard deviation of temperature (sT) isequal to 1.5 �C for smaller droplets, and0.8 �C for larger droplets.The error in the temperature measurements comes from thevariation of the vertical position of droplets smaller than thechannel height, which has an impact on the focal point and thuson the uorescence intensity.

Evaluation of dielectric and non-dielectric heating

There is an ongoing debate whether the enhancement ofchemical reactions driven by microwave irradiation is due to

Fig. 2 The temperature change of the droplets as a function of the peak-to-peakvoltage. The red curve represents droplets smaller than the channel, formed at a20 ml h�1

flow rate of benzyl alcohol. The black curve represents droplets largerthan the channel, which are compressed by the channel; they are formed at a50 ml h�1

flow rate of benzyl alcohol. The flow rate of oil is kept constant at120 ml h�1. The baseline temperature is 25 �C.


rapid heating or selective interaction of the electromagneticeld with polar molecules. Our microuidic-microwave dropletheater, due to its ability to separate conventional heat transfermechanisms from dielectric heating,41 can help shed light ontothese questions. To separate the effects of conventional heatingfrom dielectric heating, we utilize infrared imaging to visualizethe heat distribution through the device. To directly evaluatethe inuence of the carrier oil on the heat transfer from thedroplets, we measure the temperature distribution on the glasssurface of the device using an IR camera. The camera is cali-brated to determine the temperature in the images, which isreected by the color map.

We collect images at a rate of 100 Hz using an integrationtime of 2.2 ms, allowing us to monitor the change in thetemperature of the glass surface as the device heats up when apeak voltage of 8.85 V is rst applied to the microwave heater.However, what we essentially measure is the amount of heatvertically dissipated to the substrate from the droplets as themicrowave is turned on. The IR images show that the deviceheats up in the region below which the microwaves are appliedover the course of about 15 s, as illustrated in Fig. 3a. We extractthe time evolution of the lateral temperature proles of theheater and we nd that the temperature gradient is establishedon the length of 0.5 cm when the steady state is reached asshown in the inset of Fig. 3. Even aer just 10 ms the temper-ature of the glass directly below the microwave heater clearlybegins to increase; aer 500 ms we can measure the heatdissipated also in horizontal directions. Steady state conditionsare established aer 15 s and the temperature of the deviceremains constant over several hours of continuous operation.This temperature gradient shows that the benzyl alcohol

Fig. 3 IR image sequence of microwave heating taken at 100 frames per second.Inset bottom, right – the temperature profile as a function of distance from thecenter of the channel.

Nanoscale, 2013, 5, 5468–5475 | 5471

Nanoscale Paper

droplets are hotter than their surroundings, thus verifyingdielectric heating.16–18

To quantitatively determine the heat losses, we measure thetemperature of the device when it reaches steady state. Wedetermine the temperature proles along the channel from theIR images. We observe an abrupt increase in the temperature atthe entrance of the microwave heater and a correspondingabrupt decrease at the exit; the temperature remains constantalong the heated area. Adjusting the applied microwave voltagefrom 0 to 13.2 V peak-to-peak changes the temperature from 25to 45 �C as shown in Fig. 4a and b. This represents an increaseof 20 �C, somewhat less than that of the drops themselves. Todifferentiate microwave heating of the uid from resistiveheating we remove the benzyl alcohol from the device and owonly non-absorbing uorinated oil (FC-40). In this case, there isvery little temperature change measured, as the device onlyheats up by amaximum of 5 �C as shown in Fig. 4c; this residualtemperature rise can be attributed to traces of resistive heating,which may come from artifacts such as electrode imperfections.

Honeycomb-like microstructure made of nanoparticles

The ability to rapidly deliver power to the non-aqueous dropletsusing dielectric heating offers new opportunities to synthesize

Fig. 4 (a) Two-dimensional temperature map of heat dissipation at the surfaceof a microfluidic-microwave heater. In the inset is shown an optical image of thedroplets entering the microwave heater area, (b) temperature profiles along thechannel at different peak-to-peak voltages in the presence of microwave-absorbing benzyl alcohol droplets, and (c) corresponding background tempera-ture profiles when only the non-absorbing FC-40 oil is flowing.

5472 | Nanoscale, 2013, 5, 5468–5475

inorganic materials. To illustrate this potential, we use ourmicrouidic microwave heater to synthesize tungsten oxidenanoparticles within benzyl alcohol droplets using thesynthesis protocol for a conventional reaction in oil bathdescribed by Niederberger et al.24 We use the microwave-microuidic device described above to generate droplets oftungsten hexachloride and benzyl alcohol solution in FC-40 oil.The droplets of ca. 50 mm reside in the area exposed to dielectricheating of 50 �C for 64 ms. An example of droplets and theirassembly aer evaporation of FC-40 is shown in Fig. 5. Aerdrying in air, and without further washing, the unique honey-comb-like microstructure is formed. If the structure is driedonto the polished Si-wafer in a vacuum, the honeycomb-likemicrostructure is only partially preserved. Most likely the pooradhesion between the dried droplets and the Si substrate resultsin the partial exfoliation of the microstructure under vacuum(Fig. 6a). A closer look at the dried droplets shows that there ismore solid phase collected at the boundaries between thedroplets than in the middle of the single droplet. Additionally, awrinkled microstructure inside the droplets is observed(Fig. 6b). TEM images not only conrm the morphology of thedried droplets but also reveal a subtle nanostructure. Thedarker regions inside the droplets, clearly visible already in theSEM image as wrinkles, and the edges of the droplets consist ofassemblies of primary nanoparticles (Fig. 7a–c).

Inside the droplets the individual nanoparticles areobserved. HR-TEM images deliver the nal proofs of theircrystallinity (Fig. 7d–f). The size of primary particles does notexceed 3 nm. The XRD pattern shows broad reections typical

Fig. 5 Optical microscopy images of a single monolayer of benzyl alcoholdroplets: (a) benzyl alcohol droplets immerse in the continuous FC-40 phase selfassemble onto a glass surface, (b) after the continuous phase (FC-40) evaporatesthe droplets are still stable and adapt a honeycomb-like microstructure.


Fig. 6 SEM images of dried droplets at different magnifications, which show (a)a partially preserved honeycomb-like microstructure, and (b) wrinkle-like featuresinside the droplets.

Fig. 7 TEM and HR-TEM images of dried droplets. Green arrow points the spot, whicdo not exhibit preferential crystallographic orientation.


Paper Nanoscale

for nanoparticles that complicates an unambiguous identica-tion of the crystal structure. However, the best match amongdifferent tungsten oxide hydrates (WO3$nH2O, n¼ 0, 0.33, 0.5, 1and 2) is obtained with orthorhombic tungsten oxide 0.3-hydrate, ICDD Nr. 01-072-0199 (Fig. 8). Interestingly, the size ofnanoparticles and their structure change if they are washedwith ethanol, which can be observed in the sharp reection inthe XRD pattern (Fig. 8b). Also in the HR-TEM image individual2 nm nanoparticles cannot be distinguished anymore, butrather nanoparticles of 10 nm and their agglomerates (ESIFig. S7†). The diffraction rings in the corresponding electrondiffraction (ED) pattern are characteristic of polycrystallinematerials. The sharp reections in XRD and SEAD patternscan be attributed to the tungstate oxide hydroxide nano-particles (Fig. 8b).

It is likely that benzyl alcohol not only acts as a reactionmedium, but also as a capping agent. By washing with ethanol,the organic residuals are removed from the surface of tungstenoxide hydrates. We believe the growth process is driven by thedecrease in total system energy by reducing the surface energy.Furthermore, heating by irradiation with an electron beamaccelerates nanoparticle transformation and growth. This isvisualized by consecutive recording of the diffraction patterns atthe same spot. We observe a successive transformation of thediffraction rings (ESI Fig. 7a†) into well-developed lattice planes

h is magnified in the following image (from a to c). The nanoparticle agglomerates

Nanoscale, 2013, 5, 5468–5475 | 5473

Fig. 8 XRD patterns of the reaction product: (a) without washing, (b) afterwashing with ethanol. The sharp reflections are due to anisotropic growth.Vertical green bars at the bottom: reference pattern of WO3$0.33H2O, ICDDno. 01-072-0199.

Nanoscale Paper

(ESI Fig. 8†). It is not surprising since benzyl alcohol is known tostabilize small nanoparticles and metastable crystal struc-tures.32,42 For instance, we have shown that benzyl alcohol canstabilize 1–2 nm large MoO2 nanoparticles in the hexagonalcrystal structure. If acetophenone is added to the reactionsolution, the nanoparticles grow through oriented attachmentand transform into the thermodynamically more stable mono-clinic phase. Also the role of electron beam irradiation inoriented attachment and growth of nanoparticles was recentlyvisualized by real-time HR-TEM imaging of the growth of Pt3Feand growth of Au nanoparticles at the surface of CdSe.43,44 In thecase of Au on CdSe Meyns et al. showed that during irradiationof the sample by the electron beam, the thin shell of Au at thesurface of CdSe nanoparticles evolved into dot-like depositswith high contrast.44 The authors suggest that it can be corre-lated with the lower degree of coverage with ligands at thesurface of CdSe. Whereas in the case of Pt3Fe the observationsrevealed the growth of winding polycrystalline nanoparticlechains by shape-directed nanoparticle attachment followed bystraightening and orientation and shape corrections to yieldsingle-crystal nanorods.43

Finally, we performed the control bulk experiments in thelaboratory microwave reactor at 60 �C and 120 �C (see ESIFig. S9†). The XRD pattern of nanoparticles synthesized at120 �C over 10 minutes can be clearly assigned to theWO3$H2O crystal structure (ICDD no. 00-018-1418). At 60 �C wedid not observe nanoparticle formation at reaction timesshorter than 15 minutes. Even though the corresponding XRDpattern shows very broad reections it can also be assigned toWO3$H2O.

The nanoparticles synthesized in the microwave-micro-uidic device adopt the WO3$0.33H2O crystal structure incontrast to WO3$H2O obtained in our control synthesis in thelaboratory microwave reactor and experiments in oil bath.24,25 Atthe same time the reaction time is decreased from 15minutes at

5474 | Nanoscale, 2013, 5, 5468–5475

60 �C in a bulk laboratory microwave reactor down to only 64msin a microuidic device, demonstrating the advantages ofthis approach. While most previous studies focus on the engi-neering of the tungsten oxide nanoparticles, wires and theirsubsequent assembly,27–29 our technique offers a route tosimultaneously synthesize and assemble the nanoparticles.Recently, a versatile and simple approach to produce tailor-made hierarchical porous materials was presented.45 Using amicrouidic device to produce monodisperse templatingdroplets of tunable size, the materials with up to three levels ofhierarchy were prepared. Adding the microwave-reactor to themicrouidic chip could additionally give a control on thechemical composition of porous structures.

Conclusions

Although microuidic reactors are oen applied to synthesizecrystalline nanoparticles, microwave heating has not previ-ously been utilized. Here we show that with microwave heatingof pL-sized droplets we can produce appropriate conditions tocrystallize inorganic metal oxide nanoparticles. This techniqueenables future work to alter the concentration of the precursorand the heating time of the droplets, to nely control theassembly of nanoparticles within each droplet. The use of ourmicrouidic chip leads to drastically reduced synthesis time(from 15 minutes in a bulk microwave reactor, down to 64 msin a microuidic device). Additionally, our chip has thepotential to help address the important issue of distinguishingthe effects of non-thermal and thermal effects of microwaveirradiation in microwave driven particle synthesis.19 The twomain constraints in addressing this question are: the limitedcompatibility of the standard laboratory microwave reactorwith spectroscopic equipment; and the low temporal resolu-tion of most of the available spectroscopic techniques. Withthe help of single droplet uorescence detection and infraredthermal imaging techniques we can ensure that heat isgenerated only due to dielectric heating of benzyl alcohol.Therefore, in future, a droplet-based microuidic system canprovide a valuable platform for coupling of a dielectric heatingand in situ spectroscopy techniques. Moreover, a uniqueadvantage of the application of spectroscopic techniques to amicrouidic chip is the spatial resolution, which can betranslated into the enhanced temporal resolution without lossof data quality.46,47 Hence, these results reported help assessthe utility of microuidic microwave heating for exceptionallyfast inorganic material synthesis.

Acknowledgements

This work was supported by the NSF (DMR-1006546) theHarvard MRSEC (DMR-0820484), the Swiss National Founda-tion, projects: PA00P2 129086 and 2-77354-12 and by theElectron Microscopy of ETH Zurich (EMEZ). Martin Suess(EMEZ) help with HR-TEM and Rahel Boehlen support withthe so-litography are greatly acknowledged. DK thanksProf. Markus Niederberger (ETH Zurich) for his enthusiasticsupport.


Paper Nanoscale

Notes and references

1 A. P. Alivisatos, Science, 1996, 271, 933–937.2 J. Il Park, A. Saffari, S. Kumar, A. Gunther and E. Kumacheva,Annu. Rev. Mater. Res., 2010, 40, 415–443.

3 A. J. deMello, Nature, 2006, 442, 394–402.4 A. Abou-Hassan, O. Sandre and V. Cabuil, Angew. Chem., Int.Ed., 2010, 49, 6268–6286.

5 E. M. Chan, A. P. Alivisatos and R. A. Mathies, J. Am. Chem.Soc., 2005, 127, 13854–13861.

6 S.-A. Leung, R. F. Winkle, R. C. R. Wootton and A. J. deMello,Analyst, 2005, 130, 46–51.

7 I. Shestopalov, J. D. Tice and R. F. Ismagilov, Lab Chip, 2004,4, 316–321.

8 Y. Song, H. Modrow, L. L. Henry, C. K. Saw, E. E. Doomes,V. Palshin, J. Hormes and C. S. S. R. Kumar, Chem. Mater.,2006, 18, 2817–2827.

9 S. Xu, Z. H. Nie, M. Seo, P. Lewis, E. Kumacheva, H. A. Stone,P. Garstecki, D. B. Weibel, I. Gitlin and G. M. Whitesides,Angew. Chem., Int. Ed., 2005, 44, 724–728.

10 A. Aimable, N. Jongen, A. Testino, M. Donnet, J. Lemaitre,H. Hofmann and P. Bowen, Chem. Eng. Technol., 2011, 34,344–352.

11 W. Ehrfeld, V. Hessel and H. Lowe, Microreactors: NewTechnology for Modern Chemistry, Wiley-VCH, 2000.

12 A. M. Nightingale, S. H. Krishnadasan, D. Berhanu, X. Niu,C. Drury, R. McIntyre, E. Valsami-Jones and J. C. deMello,Lab Chip, 2011, 11, 1221–1227.

13 B. F. Cottam, S. Krishnadasan, A. J. deMello, J. C. deMelloand M. S. P. Shaffer, Lab Chip, 2007, 7, 167–169.

14 Z. L. Xue, A. D. Terepka and Y. Hong, Nano Lett., 2004, 4,2227–2232.

15 I. Bilecka and M. Niederberger, Nanoscale, 2010, 2, 1358–1374.

16 D. Issadore, K. J. Humphry, K. A. Brown, L. Sandberg,D. A. Weitz and R. M. Westervelt, Lab Chip, 2009, 9, 1701–1706.

17 J. Geist, J. J. Shah, M. V. Rao and M. Gaitan, J. Res. Natl. Inst.Stand. Technol., 2007, 112, 177–189.

18 J. J. Shah, S. G. Sundaresan, J. Geist, D. R. Reyes, J. C. Booth,M. V. Rao and M. Gaitan, J. Micromech. Microeng., 2007, 17,2224–2230.

19 S. A. Galema, Chem. Soc. Rev., 1997, 26, 233–238.20 B. Gestblom, Chem. Phys. Lett., 1980, 74, 333–336.21 R. Harpeness and A. Gedanken, Langmuir, 2004, 20, 3431–

3434.22 X. Hu, J. Gong, L. Zhang and J. C. Yu, Adv. Mater., 2008, 20,

4845–4850.23 M. Tsuji, M. Hashimoto, Y. Nishizawa, M. Kubokawa and

T. Tsuji, Chem.–Eur. J., 2005, 11, 440–452.24 M. Niederberger, M. H. Bartl and G. D. Stucky, J. Am. Chem.

Soc., 2002, 124, 13642–13643.25 I. Olliges-Stadler, J. Stotzel, D. Koziej, M. D. Rossell,

J.-D. Grunwaldt, M. Nachtegaal, R. Frahm andM. Niederberger, Chem.–Eur. J., 2012, 18, 2305–2312.


26 J. Polleux, A. Gurlo, N. Barsan, U. Weimar, M. Antonietti andM. Niederberger, Angew. Chem., Int. Ed., 2006, 45, 261–265.

27 H. Zheng, J. Z. Ou, M. S. Strano, R. B. Kaner, A. Mitchell andK. Kalantar-Zadeh, Adv. Funct. Mater., 2011, 21, 2175–2196.

28 M. Breedon, P. Spizzirri, M. Taylor, J. du Plessis,D. McCulloch, J. Zhu, L. Yu, Z. Hu, C. Rix, W. Wlodarskiand K. Kalantar-zadeh, Cryst. Growth Des., 2010, 10, 430–439.

29 K. Kalantar-zadeh, A. Vijayaraghavan, M.-H. Ham, H. Zheng,M. Breedon and M. S. Strano, Chem. Mater., 2010, 22, 5660–5666.

30 A. C. Siegel, S. S. Shevkoplyas, D. B. Weibel, D. A. Bruzewicz,A. W. Martinez and G. M. Whitesides, Angew. Chem., Int. Ed.,2006, 45, 6877–6882.

31 C. Holtze, A. C. Rowat, J. J. Agresti, J. B. Hutchison,F. E. Angile, C. H. J. Schmitz, S. Koster, H. Duan,K. J. Humphry, R. A. Scanga, J. S. Johnson, D. Pisignanoand D. A. Weitz, Lab Chip, 2008, 8, 1632–1639.

32 G. Garnweitner and M. Niederberger, J. Mater. Chem., 2008,18, 1171–1182.

33 K. G. Casey and E. L. Quitevis, J. Phys. Chem., 1988, 92, 6590–6594.

34 J. F. Lou, T. M. Finegan, P. Mohsen, T. A. Hatton andP. E. Laibinis, Rev. Anal. Chem., 1999, 18, 235–284.

35 S. Ebert, K. Travis, B. Lincoln and J. Guck, Opt. Express, 2007,15, 15493–15499.

36 J. Sakakibara and R. J. Adrian, Exp. Fluids, 1999, 26, 7–15.37 T. Robinson, Y. Schaerli, R. Wootton, F. Hollfelder,

C. Dunsby, G. Baldwin, M. Neil, P. French and A. deMello,Lab Chip, 2009, 9, 3437–3441.

38 D. Ross, M. Gaitan and L. E. Locascio, Anal. Chem., 2001, 73,4117–4123.

39 R. Samy, T. Glawdel and C. L. Ren, Anal. Chem., 2008, 80,369–375.

40 3M� Fluorinert� Electronic Liquid FC-40, http://www.solutions.3m.com, 2011.

41 D. Obermayer, B. Gutmann and C. O. Kappe, Angew. Chem.,Int. Ed., 2009, 48, 8321–8324.

42 D. Koziej, M. D. Rossell, B. Ludi, A. Hintennach, P. Novak,J.-D. Grunwaldt and M. Niederberger, Small, 2011, 7, 377–387.

43 H.-G. Liao, L. Cui, S. Whitelam and H. Zheng, Science, 2012,336, 1011–1014.

44 M. Meyns, N. G. Bastus, Y. Cai, A. Kornowski, B. H. Juarez,H. Weller and C. Klinke, J. Mater. Chem., 2010, 20, 10602–10605.

45 A. R. Studart, J. Studer, L. Xu, K. Yoon, H. C. Shum andD. A. Weitz, Langmuir, 2011, 27, 955–964.

46 K. L. A. Chan, X. Niu, A. J. de Mello and S. G. Kazarian, LabChip, 2010, 10, 2170–2174.

47 H. Oyanagi, Z. H. Sun, Y. Jiang, M. Uehara, H. Nakamura,K. Yamashita, L. Zhang, C. Lee, A. Fukano and H. Maeda,J. Synchrotron Radiat., 2011, 18, 272–279.

Nanoscale, 2013, 5, 5468–5475 | 5475

Nanoscale - Harvard University

Documents

Transcript of Nanoscale - Harvard University